feat(lake): publish gpci_proposed + primary_care_service_code to DuckLake (closes #630)

dev/scripts/publish_reference_to_lake.py

@@ -26,6 +26,7 @@ SCHEMAS: dict[str, tuple[str, ...]] = {
         "carrier_locality",
         "clinical_labor",
         "gpci",
+        "gpci_proposed",
         "medical_equipment",
         "medical_supply",
         "physician_work_time",
@@ -36,6 +37,19 @@ SCHEMAS: dict[str, tuple[str, ...]] = {
     "cms": ("ingest_log",),
 }
 
+# (monolith source, lake destination) — tables whose lake name differs
+# from their monolith name. The MSSP designated-primary-care-service
+# list (42 CFR 425.400(c), used for ACO beneficiary assignment) lives
+# monolith-side as a tuva/dbt intermediate; publish it under a stable
+# reference name the notebooks can cite (#630). Published whenever the
+# destination's schema is selected via --schemas.
+RENAMED: tuple[tuple[str, str], ...] = (
+    (
+        "cms_provider_attribution._primary_care_hcpcs_codes",
+        "cms.primary_care_service_code",
+    ),
+)
+
 
 def main() -> int:
     parser = argparse.ArgumentParser(description=__doc__)
@@ -66,18 +80,27 @@ def main() -> int:
     )
 
     src = connect.duckdb("aco")  # read-only; resolves to the replica
+
+    def _publish(source: str, dest: str) -> bool:
+        t = time.time()
+        df = src.execute(f"SELECT * FROM {source}").pl()  # noqa: S608
+        ctx.save(dest, nw.from_native(df), mode="replace")
+        back = nw.to_native(ctx.load(dest))
+        ok = back.height == df.height
+        label = source if source == dest else f"{source} → {dest}"
+        print(
+            f"  {label}: {df.height} rows → lake "
+            f"({back.height} read back) [{'OK' if ok else 'MISMATCH'}] "
+            f"in {time.time() - t:.1f}s"
+        )
+        return ok
+
     for schema in args.schemas:
         for table in SCHEMAS[schema]:
-            t = time.time()
+            if not _publish(f"{schema}.{table}", f"{schema}.{table}"):
-            df = src.execute(f"SELECT * FROM {schema}.{table}").pl()  # noqa: S608
+                return 1
-            ctx.save(f"{schema}.{table}", nw.from_native(df), mode="replace")
+    for source, dest in RENAMED:
-            back = nw.to_native(ctx.load(f"{schema}.{table}"))
+        if dest.split(".", 1)[0] in args.schemas and not _publish(source, dest):
-            status = "OK" if back.height == df.height else "MISMATCH"
-            print(
-                f"  {schema}.{table}: {df.height} rows → lake "
-                f"({back.height} read back) [{status}] in {time.time() - t:.1f}s"
-            )
-            if status != "OK":
             return 1
     src.close()
     print("lake publish complete")